sicbogame| Build the mysterious Zurich team! Apple is staring at Google for "robbery": It is said that it has poached 36 AI talents

Financial Associated Press, April 30 (editor Xiaoxiang) since Apple gave up a decade of car-building efforts earlier this yearSicbogameSince the decision to fully deploy artificial intelligence (AI), the progress of its artificial intelligence strategy has attracted much attention from the outside world.

According to a recent media release, Apple has put a lot of effort into the AI field in recent years-poaching dozens of artificial intelligence experts from Google over the past six years and creating a mysterious European laboratory in Zurich. The technology giant is forming a team to compete with competitors in developing new artificial intelligence models and products.

According to media analysis of hundreds of LinkedIn profiles, public recruitment messages and research papers, Apple has actually launched a hiring spree in recent years to expand its global artificial intelligence and machine learning team.

The iPhone maker, in particular, is targeting employees from Google. According to statistics, at least 36 experts have been poached from this competitor since John Giannandrea was poached as Apple's senior vice president for machine learning and AI strategy in 2018.

By contrast, Apple derives from itsSicbogameThe number of AI talents poached by his company is basically controlled in single digits.

Although most of Apple's artificial intelligence team works in offices in California and Seattle, the tech giant has also expanded an important outpost laboratory in Zurich.

Luc Van Gool, a professor at the Federal Institute of Technology in Zurich, said Apple's acquisition of two local artificial intelligence start-ups, VR FaceShift and image recognition company Fashwell, enabled Apple to set up a research laboratory in Zurich, the mysterious Zurich Vision Lab that has been widely reported in recent years.

It is reported that Apple employees in Zurich participated in Apple's research on the underlying technology that supports ChatGPT. Their papers focus on more and more advanced artificial intelligence models that combine text and visual input to respond to a variety of queries.

The company has been posting job ads for generative artificial intelligence (AIGC) at two locations in Zurich, one of which is particularly low-key. According to media visits, many of the "neighbors" who are located near Apple's offices do not even know that the office exists.

Apple layout AI actually has a long history.

Although Apple's main competitors in AI, such as Microsoft, Google and Amazon, have been bragging about their multibillion-dollar investments in the cutting-edge technology of artificial intelligence in recent years, Apple has long been tight-lipped about its artificial intelligence plans.

The lack of any eye-catching AI products so far has put Apple's share price under pressure this year.

However, Apple has been developing artificial intelligence products for more than a decade, such as its voice assistant Siri. The company has long recognized the potential of "neural networks"-a form of artificial intelligence inspired by the way neurons interact in the human brain and a technology behind breakthroughs such as ChatGPT.

Chuck Wooters, an expert on interactive artificial intelligence and large language models, joined Apple in December 2013 and worked on the Siri team for nearly two years. "while I was working there, one of the things that the Siri team was pushing was to shift speech recognition to neural architecture," Wooters said. Even at that time, before the rise of large language models, they were already loyal supporters of neural networks. "

This interest seems to have led Apple to crave neural network researchers for a long time.

In 2016, Apple acquired Perceptual Machines, a company founded by Ruslan Salakhutdinov and two of his students at Carnegie Mellon University, dedicated to post-graduate AI-driven image detection. "around that time, they were looking for researchers and trying to build the infrastructure to train these models," Salakhutdinov said in an interview with the media. "

Salakhutdinov is a key figure in the history of neural networks. He studied under the technology's "godfather" Geoffrey Hinton at the University of Toronto. Salakhutdinov, a former director of artificial intelligence research at Apple, did not return to Carnegie Mellon University until 2020 to return to academia.

Is Apple's AI team full of Google celebrities?

It is reported that Apple's top artificial intelligence team is now mostly made up of Google's former core figures, including the famous John Giannandrea, who led the Google Google Brain team, the early artificial intelligence laboratory of Google that was later merged by DeepMind.

Samy Bengio, Apple's senior director of artificial intelligence and machine learning research, was once one of Google's top AI scientists. Ruoming Pang, who led Apple's Foundation Models team on large language models, led Google's AI speech recognition research.

Apple also hired another machine deep learning pioneer, Ian Goodfellow, but he returned to Google in 2022 to protest Apple's return policy during the epidemic.

In an important research paper published by Apple in March, six former Google employees who had been employed by Apple in the past two years were listed as signed authors. Apple revealed in its paper that it has developed a series of artificial intelligence models called "MM1" that use text and visual input to generate responses.

According to many industry insiders, Apple may now be focusing on deploying AIGC on its mobile devices, which would be a breakthrough, allowing artificial intelligence chatbots and applications to run on the phone's own hardware and software, rather than the computing power provided by the cloud in the data center.

Salakhutdinov said Apple has been focused on "doing as much as possible on the device", which will lead to the need for more powerful chips with so-called dynamic random access memories (DRAM) that can handle the large amounts of data needed to power artificial intelligence models.

Sumit Sadana, executive vice president and chief business officer of Micron Technology, one of Apple's chip suppliers, said, "the next big event will be AI smartphones-these phones will need more DRAM. At present, the average capacity of a smartphone memory chip is about 8GB, but at least 12GB is needed to run a large language model. "

Industry insiders expect that Apple's new attempt at AI technology may make its debut at the World Developers Conference (WWDC) in June.

Morgan Stanley analyst Erik Woodring said the next iPhone "may be more like a voice-activated smart personal assistant, led by an upgraded version of Siri that can interact with all applications on the phone through voice controls, for example."